Some animals are ready to go as soon as they are born. These are called precocial animals. They are born knowing how to walk.
It's interesting seeing what comes built-in. You can see this if you watch a horse being born. Within the first hour, the foal will stand, and despite long legs, this usually works the first time. Lying down, however, is not preprogrammed. I've watched a foal circle trying to figure out how to get down from standing, and finally collapsing to the ground in a heap. Standing up quickly is essential to survival, but smoothly lying down is not.
Within a day, a newborn foal can run with the herd.
Of the mammals, most of the equines and some of the rodents (beavers) are precocial.
Pigs are, monkeys are not. It's not closely tied to evolutionary ancestry.
You forgot to mention the most important part: Those animals are the norm, it's us humans who are the outliers! Of course we have remnants of this knowledge and models of the world that formed our ancestors DNA for Millions of yers in us. Our ancestors newborns were as independent as these other animal children.
The only difference is that our heads are too big, so we are born prematurely. The term precocial is misleading. We aren't the norm, we are the outliers, human newborn's are not a blank plate, basing your models of intelligence and learning on the assumption we are the best fit for survival in a world is bogus.
Most animals know how to walk. They have pre-build 'knowledge' and start with it when they get muscles. The main difference is some species develop muscles before they are borne. Others don't, some can't even see. But as soon as they develop they can walk without even seeing how others do it. The same way birds can learn, or be trained to fly without seeing examples. They start flapping both wings in sync, this is pre-build.
Wow! Some years ago I was thinking about reasons for why people on ADHD/autism spectrum are different.
First heard somewhere (don't remember where or exact idea) that neurons initially form groups and these groups then perform functions. This led to an idea that if someone's brain sacrificed some "copy other primate" groups for "pattern recognition" groups, you would get a unit with higher IQ for non social use, without changing the brain to be more effective in general. This would come at a cost to social/copying skills. This idea doesn't explain "systems thinking" tendency or "not seeing forest for the trees" tendency in autist spectrum folks.
On another occasion, it occurred to me that regular brain run / loop consists of a short reality check and longer flow state. If there are too many reality checks, you get anxiety and can't work effectively. OTOH, too little realty checks and you get stuck on non important things. At the same time, impairing this "check to flow" balance in a safe (non anxiety provoking) environment would result in an individual that could perform the kind of deeper work with results not achievable by not modified individuals.
Have watched 50+ h of psychology lectures, but don't have any formal knowledge on these things so please take it with a grain of salt.
Edit: myself I'm formally on ADHD, and in personal opinion also on Autism spectrum. Just learned to "act normal" very well by the time I got into diagnosis.
How newborn brain works is absolutely fascinating for me. I just don't understand how is it possible.
Human DNA contains 1.5 GB information.
Human body, including brain, gets built using this information only. So our "preconfigured" neural networks are also built using this information only.
And apparently it's enough to encode complex behaviour. That's not just visible things. Brain processes a humongous amount of information, it basically supports living processes for entire body, processing miriads of sensors, adjusting all kinds of knobs for body to function properly.
I just don't understand how is it possible just from a purely bit size approach. For me, it's a mystery.
I'll probably get downvoted for this, but recent (last ~100 years) evidence and phenomena suggest that consciousness might be fundamental to reality, and thus there could be some other information transfer we would currently consider "woo" going on here. This is hard (if not impossible) to prove, of course, but quantum mechanics has totally bewildered many aspects of the materialist ("reductionist") model of the universe. There is a large and increasing number of physicists and other reputed scientists/researchers who are adopting some variation of the consciousness-as-fundamental stance.
This seems like it was proven ages ago with the no-free-lunch theorem.
Humans could not learn to function unless their brains encode a useful prior for learning about the world. That prior means "preconfigured with instructions for understanding the world".
The short form of the no-free-lunch theorem is that if there is no prior (i.e. all possible universes are equally likely) then for any learning problem P there are an equal number of universes that learning system A will outdo learning system B on that problem.
If not all universes are equally likely, one learning system can vastly outdo another or even most other learning systems. Not equally likely is the assumption built into brains. Without that, you can't learn effectively.
So the biology is just implementation of that general principle. The details of how that implementation works are interesting, but whether we are preconfigured for learning was never in question.
Maybe I’m overlooking something, but wouldn’t this be similar to an instinct that is preprogrammed from natural selection? For example, sea turtles know they need to move from the beach toward the ocean, and spiders know how to spin their species specific web pattern. No-one teaches the sea turtles or spiders how to do this. Wouldn't this be the same for our thoughts and thinking?
I've always found Retinal Waves[1] interesting. During development of the visual system, there are spontaneous bursts of activity without external stimuli, helping the synapses to organize properly.
In my layman's view, it's like hallucinating shapes that are important to learn. Very similar to the "priming" described in the article, but easier to visualize (literally).
I don't know if it's comedy or tragedy, but I've often considered the situation of what a newborn is "expecting".
"Ok, first thing when I come out is I'm gonna meet the family. I'll try to get used to their face, whoever I see first. Maybe they'll show me around the savannah, should be a lot of sunshine, colours, blue sky. Then I'll sleep directly on my mom and get some boob milk."
Kid comes out, everyone is wearing a mask, half of them aren't family, they're indoors with artificial light, and they have "clothes" put on them, and are put in a cot to sleep.
There’s a difference between reacting to your environment, and being intelligent. Many top scientists working in AI, I know from firsthand experience, are disconnected from how any form of responsiveness appears in nature. It’s asinine to try to build or recreate something you don’t understand, why even bother?
No. The headline does not match the justification in the article. The organoid brain tissue is not hooked up to sensory mechanisms in its first months, I accept that, but they are under the influence of input-output training in their initial structure which would reasonably form some non-random pattern of weights, due to characteristics of the cells.
The question then is, 1) are these characteristics acting as some kind of evolutionary adaption that passes on preconfigured world recognition (asserted by the headline), 2) are they some kind of evolutionary adaptation that makes more effective thinking systems in the form of some specific cognitive structure (more likely IMO), ie they are random features that cause non-random neural structure that drive survival-selection.
Think about the process for (1) to occur. Some ancestor learned in their life to fear snake-like animals or crave mama’s nearness, what possible process could put that knowledge (neural structure of such specificity) into the way that animal generated its sperm or egg? On the other hand, it’s reasonable to assume some genetic encodings encourage specific neural structures even in very early stages, that these are random, but that evolution favored some vs others over the 500mm years animals-with-brains have been around.
There was quite an interesting discussion with Hinton explaining to Jon Stewart how he thinks of brain function, kind of anthropomorphising neurons as things that can see patterns and all they can do is go ping when they see one (https://youtu.be/jrK3PsD3APk?t=366) So if one in the retina say pings if it sees a line and then one behind might see a pattern of a horizontal line ping and a vertical line ping and ping for a cross.
Anyway I was thinking for that to work the neurons would have to kind of chat to each other like "here I am, who's receiving me" etc. Also some communication that if you are differentiating say crosses and circles, the cross neuron can say "hey I've got this one" so the other can go "ok, I'll do the circle then" so the neurons differentiate to recognize different things.
I guess some of that sort of communication system maybe goes on before there is sensory input so the neurons kind of know how they are wired?
One difference with the Hinton/Stewart talk is there he was saying all they can do is go ping, whereas the article has "firing off a complex repertoire of time-based patterns, or sequences" which makes sense - you'd have a job sorting it with simple pings.
66 comments
[ 2.6 ms ] story [ 76.3 ms ] threadIt's interesting seeing what comes built-in. You can see this if you watch a horse being born. Within the first hour, the foal will stand, and despite long legs, this usually works the first time. Lying down, however, is not preprogrammed. I've watched a foal circle trying to figure out how to get down from standing, and finally collapsing to the ground in a heap. Standing up quickly is essential to survival, but smoothly lying down is not. Within a day, a newborn foal can run with the herd.
Of the mammals, most of the equines and some of the rodents (beavers) are precocial. Pigs are, monkeys are not. It's not closely tied to evolutionary ancestry.
The only difference is that our heads are too big, so we are born prematurely. The term precocial is misleading. We aren't the norm, we are the outliers, human newborn's are not a blank plate, basing your models of intelligence and learning on the assumption we are the best fit for survival in a world is bogus.
Most animals know how to walk. They have pre-build 'knowledge' and start with it when they get muscles. The main difference is some species develop muscles before they are borne. Others don't, some can't even see. But as soon as they develop they can walk without even seeing how others do it. The same way birds can learn, or be trained to fly without seeing examples. They start flapping both wings in sync, this is pre-build.
who is doing it? why the observed instructions are chosen?
First heard somewhere (don't remember where or exact idea) that neurons initially form groups and these groups then perform functions. This led to an idea that if someone's brain sacrificed some "copy other primate" groups for "pattern recognition" groups, you would get a unit with higher IQ for non social use, without changing the brain to be more effective in general. This would come at a cost to social/copying skills. This idea doesn't explain "systems thinking" tendency or "not seeing forest for the trees" tendency in autist spectrum folks.
On another occasion, it occurred to me that regular brain run / loop consists of a short reality check and longer flow state. If there are too many reality checks, you get anxiety and can't work effectively. OTOH, too little realty checks and you get stuck on non important things. At the same time, impairing this "check to flow" balance in a safe (non anxiety provoking) environment would result in an individual that could perform the kind of deeper work with results not achievable by not modified individuals.
Have watched 50+ h of psychology lectures, but don't have any formal knowledge on these things so please take it with a grain of salt.
Edit: myself I'm formally on ADHD, and in personal opinion also on Autism spectrum. Just learned to "act normal" very well by the time I got into diagnosis.
Human DNA contains 1.5 GB information.
Human body, including brain, gets built using this information only. So our "preconfigured" neural networks are also built using this information only.
And apparently it's enough to encode complex behaviour. That's not just visible things. Brain processes a humongous amount of information, it basically supports living processes for entire body, processing miriads of sensors, adjusting all kinds of knobs for body to function properly.
I just don't understand how is it possible just from a purely bit size approach. For me, it's a mystery.
Humans could not learn to function unless their brains encode a useful prior for learning about the world. That prior means "preconfigured with instructions for understanding the world".
The short form of the no-free-lunch theorem is that if there is no prior (i.e. all possible universes are equally likely) then for any learning problem P there are an equal number of universes that learning system A will outdo learning system B on that problem.
If not all universes are equally likely, one learning system can vastly outdo another or even most other learning systems. Not equally likely is the assumption built into brains. Without that, you can't learn effectively.
So the biology is just implementation of that general principle. The details of how that implementation works are interesting, but whether we are preconfigured for learning was never in question.
In my layman's view, it's like hallucinating shapes that are important to learn. Very similar to the "priming" described in the article, but easier to visualize (literally).
[1] https://en.wikipedia.org/wiki/Retinal_waves
"Ok, first thing when I come out is I'm gonna meet the family. I'll try to get used to their face, whoever I see first. Maybe they'll show me around the savannah, should be a lot of sunshine, colours, blue sky. Then I'll sleep directly on my mom and get some boob milk."
Kid comes out, everyone is wearing a mask, half of them aren't family, they're indoors with artificial light, and they have "clothes" put on them, and are put in a cot to sleep.
a system prompt?
The question then is, 1) are these characteristics acting as some kind of evolutionary adaption that passes on preconfigured world recognition (asserted by the headline), 2) are they some kind of evolutionary adaptation that makes more effective thinking systems in the form of some specific cognitive structure (more likely IMO), ie they are random features that cause non-random neural structure that drive survival-selection.
Think about the process for (1) to occur. Some ancestor learned in their life to fear snake-like animals or crave mama’s nearness, what possible process could put that knowledge (neural structure of such specificity) into the way that animal generated its sperm or egg? On the other hand, it’s reasonable to assume some genetic encodings encourage specific neural structures even in very early stages, that these are random, but that evolution favored some vs others over the 500mm years animals-with-brains have been around.
Anyway I was thinking for that to work the neurons would have to kind of chat to each other like "here I am, who's receiving me" etc. Also some communication that if you are differentiating say crosses and circles, the cross neuron can say "hey I've got this one" so the other can go "ok, I'll do the circle then" so the neurons differentiate to recognize different things.
I guess some of that sort of communication system maybe goes on before there is sensory input so the neurons kind of know how they are wired?
One difference with the Hinton/Stewart talk is there he was saying all they can do is go ping, whereas the article has "firing off a complex repertoire of time-based patterns, or sequences" which makes sense - you'd have a job sorting it with simple pings.